摘要
研究了多尺度几何分析工具Contourlet,提出了一种基于层结构的Contourlet多阈值去噪算法。该算法将硬阈值算法与基于子带相关的图像去噪方法相结合,根据Contourlet变换后各层分解的系数数目及噪声强度设定阈值,并利用硬阈值函数实现图像去噪。使用该算法去噪后的图像在主观视觉效果和客观质量等方面较小波算法有显著提高。
This paper researches the multiscale geometry analysis tool Contourlet, and proposes a new Contourlet multi-threshold shrink method for image denoising. The algorithm combines hard-threshold with correlation among the suhhand layers of Contourlet transform. Thresholding is derived by both the numbers of coefficients in each transformed layer and the intensity of noises added to the original image, hard-threshold function is also adopted for image denoising. Comparing with traditional wavelet denoising methods, the algorithm achieves obvious improvement in both subjective visual effect and objective quality.
出处
《计算机工程》
EI
CAS
CSCD
北大核心
2006年第20期180-182,共3页
Computer Engineering
基金
国家自然科学基金资助项目(69975015)
教育部优秀青年教师资助计划基金资助项目
